An Integrative Analysis of the Specific Distinctness of Valvata ( Cincinna ) ambigua Westerlund, 1873 and Valvata ( Cincinna ) piscinalis (Müller, 1774) (Gastropoda: Valvatidae)

: Valvata ( Cincinna ) piscinalis (Müller, 1774) is a widespread and variable Palaearctic freshwater snail species. Some authors have separated more depressed forms with a wider umbilicus as a distinct species, Valvata ( Cincinna ) ambigua Westerlund, 1873. The latter species was described from Scandinavia and has also been reported from Siberia and Kazakhstan and more recently from Central Europe. We conducted an integrative study of the delimitation and relationships of V. ambigua and V. piscinalis using both morphometric and molecular genetic analyses. Analyses of the morphometric data did not reveal differentiation into distinct clusters. Rather, the shell characteristics used to distinguish V. ambigua and V. piscinalis showed continuous variation. There is little variability in mitochondrial DNA sequences in the V. piscinalis complex. A median-joining network based on cytochrome oxidase sequences showed that the morphological character states supposedly characteristic of V. ambigua and V. piscinalis did not correlate with the genetic relationships of the individuals studied. We therefore consider V. ambigua to be synonymous with V. piscinalis .


Introduction
Valvatidae Gray, 1840 (Gastropoda: Heterobranchia) are a group of freshwater snails from the Northern Hemisphere.They are most diverse in the Palaearctic, with approximately 60 species, while 10 species are known from the Nearctic and only one from the Afrotropical region [1,2].However, many of the species are in need of revision.Even for some European species, the delimitation is still unclear.While Western European malacologists distinguished only five nominal species of the genus Valvata Müller, 1773 in Northern and Central Europe [3,4], Russian and Ukrainian researchers recognized up to 28 Valvata species in the European part of Russia and neighbouring countries alone [5,6].One of the problematic species complexes is the widespread and variable Valvata (Cincinna) piscinalis (Müller, 1774) (Figure 1a).Glöer [2] divided this species into five subspecies, four of which were later recognized as separate species by Glöer [7].In addition, Glöer [7] listed Valvata (Cincinna) ambigua Westerlund, 1873, Valvata (Cincinna) kliniensis Milachevich, 1881, and Valvata (Cincinna) lilljeborgi Westerlund, 1897, which also belong to the V. piscinalis complex, as separate species following the opinion of Anistratenko [8] and Starobogatov et al. [9].
In this study, we investigated the distinctness of V. ambigua (Figure 1c).Westerlund [10] described this species based on specimens from Göteborg in Sweden.Later, Westerlund [11] classified the nominal taxon as a variety of V. piscinalis and listed it from Sweden and Finland.In contrast, Krivosheina and Starobogatov [12] resurrected V. ambigua as a distinct species and recorded it from Western Siberia.Anistratenko [8] and Starobogatov et al. [9] also treated V. ambigua as a valid species in their comprehensive works on freshwater molluscs of Ukraine and Russia and adjacent countries, respectively.Glöer and Diversity 2024, 16, 419 2 of 11 Diercking [13] reported the species from Hamburg in Germany and discussed its ecology.Vinarski et al. [14] discussed the distinctness of V. ambigua and V. piscinalis and reported the species also from Kazakhstan and Norway.According to Glöer [7], the range of V. ambigua covers Northern, Central and Eastern Europe as well as the southern part of Western Siberia.
We used shell morphometric data as well as mitochondrial gene sequences to investigate the distinctness of V. ambigua and V. piscinalis.In particular, we studied samples from the Hamburg surroundings.This area is particularly suitable for testing the distinctness of the two nominal species because they are said to occur there syntopically [13] and because Vinarski et al. [14] also focused on material from the Hamburg region.
Diversity 2024, 16, 419 2 of 11 et al. [9] also treated V. ambigua as a valid species in their comprehensive works on freshwater molluscs of Ukraine and Russia and adjacent countries, respectively.Glöer and Diercking [13] reported the species from Hamburg in Germany and discussed its ecology.Vinarski et al. [14] discussed the distinctness of V. ambigua and V. piscinalis and reported the species also from Kazakhstan and Norway.According to Glöer [7], the range of V. ambigua covers Northern, Central and Eastern Europe as well as the southern part of Western Siberia.
We used shell morphometric data as well as mitochondrial gene sequences to investigate the distinctness of V. ambigua and V. piscinalis.In particular, we studied samples from the Hamburg surroundings.This area is particularly suitable for testing the distinctness of the two nominal species because they are said to occur there syntopically [13] and because Vinarski et al. [14] also focused on material from the Hamburg region.

Materials
Valvata specimens were collected from slowly flowing ditches and rivers in different locations in Hamburg in the summer of 2023.Geographical coordinates were recorded with a GPS and are given in WGS 84 format.The specimens were preserved in 100% isopropanol at −20 °C in the Zoological Museum Hamburg (ZMH) (Supplementary Table S1).Additional shell material from the ZMH was used for morphometric analyses.

Materials
Valvata specimens were collected from slowly flowing ditches and rivers in different locations in Hamburg in the summer of 2023.Geographical coordinates were recorded with a GPS and are given in WGS 84 format.The specimens were preserved in 100% isopropanol at −20 • C in the Zoological Museum Hamburg (ZMH) (Supplementary Table S1).Additional shell material from the ZMH was used for morphometric analyses.

Morphometric Analyses
The shell diameter (D), shell height (H), diameter of the second whorl (D2W), diameter of the umbilicus (U), diameter (da) and height (ha) of the aperture were measured using a Leica MZ9.5 stereomicroscope with an ocular micrometer (accurate to 0.05 mm).The number of whorls was counted with an accuracy of 0.25 whorls using the method described by Glöer [4].
The shells were photographed with a Passport Imaging System (Dun, Palmyra, VA, USA) with a Canon EOS 6D (Canon Inc., Tokyo, Japan) camera and processed with Zerene Stacker (Zerene Systems LLC, Richland, WA, USA) and Adobe Photoshop v. 25.9.1 (Adobe Systems Inc., San José, CA, USA) software.
We used normal mixture models (NMMs, [15]) to explore the existence of distinct morphologic clusters without a priori information about groups [16,17].Following the approach proposed by Cadena et al. [18], we first reduced the dimensionality of the phenotypic space via principal component analysis (PCA) on the covariance matrix of logtransformed data (the seven shell measurements) with the statistical software R v. 4.2.0 [19].We used the R package clustvarsel 2.3.4 [20] to select the set of principal components most useful for group discrimination.Multiple NMMs for 1-10 clusters were fitted to this multivariate dataset with the R package mclust 5.4.9 [21], and the best model was selected according to the Bayesian Information Criterion.

Alignment and Analyses of DNA Sequences
ChromasPro 2.1.10.1 (Technelysium, Tewantin, Australia) was used to assemble the forward and reverse sequence reads.The GenBank accession numbers of the new COI and 16S rDNA sequences, as well as sequences of additional Valvata and Megalovalvata Lindholm, 1906 specimens for which both the COI and 16S rRNA genes were available from the studies of Hauswald et al. [26], Dinapoli and Klussmann-Kolb [27], Clewing et al. [28] and Saito et al. [29], are listed in Supplementary Table S2.
The sequences were aligned with MAFFT v. 7.525 [30] using the Q-INS-i algorithm.Kimura 2-parameter (K2P) distances were calculated with MEGA11 [31].We selected the optimal partitioning scheme and substitution models starting with the three codon positions of cytochrome oxidase and the 16S rRNA gene as initial partitions using ModelFinder [32] implemented in IQ-TREE v. 2.2.2.6 [33].In the search for the optimal substitution models for the maximum likelihood analysis, we included the FreeRate heterogeneity model, while we restricted the search to models available in MrBayes for the Bayesian inference analysis.
Maximum likelihood phylogenetic analysis based on the concatenated sequences of the partial COI and 16S rRNA genes was performed using IQ-TREE v. 2.2.2.6 [33].We evaluated branch support with 1000 standard nonparametric bootstrap replicates.
A Bayesian inference analysis of the phylogeny was performed using MrBayes version 3.2.2[34].Metropolis-coupled Monte Carlo Markov chain searches were run with four chains in two separate runs with 10,000,000 generations.Diagnostics obtained from the MrBayes output were used to assess stationarity and convergence.The first 2,500,000 generations of each run were discarded as burn-in.
We constructed a median-joining network [35] based on COI sequences using the program POPART [36] with ε = 0.The constructed network was colour-coded according to the morphological trait ratios D/H and U/D.Vinarski et al. [14] distinguished V. piscinalis and V. ambigua based on their D/H (0.87-0.99 for V. piscinalis and 0.96-1.17for V. ambigua) and U/D ratios.We divided the overlapping range of the D/H ratio at the centre (0.975 mm) and used this value as a threshold to indicate the potential classification of individuals in the network.The corresponding threshold of U/D is based on the D/H threshold, as Vinarski et al. [14] did not measure the umbilicus.We set the threshold midway between the lowest U/D value of specimens with D/H ≥ 0.975 mm and the highest U/D value of specimens with D/H < 0.975 mm (U/D = 0.11).

Shell Morphology
We measured shell parameters of 178 specimens from several sites around Hamburg and Ratzeburg in northern Germany (Supplementary Table S1).The range of the measurements covers the range of the parameters of V. ambigua and V. piscinalis reported by Vinarski et al. [14].Plots of shell height against shell width (Figure 2a) and of relative umbilicus width (umbilicus width/shell width) against shell shape (shell width/height) (Figure 2b) show that these shell measurements and ratios vary continuously among the studied samples.This is also true if only the samples from the Reitbrooker Sammelgraben in Hamburg-Bergedorf, from which Glöer and Diercking [13] reported syntopic occurrences of V. ambigua and V. piscinalis without intermediates, were considered.The highest empirical support based on the information in principal components 4, 5 and 6 was found for two NMMs (Figure 3).However, one NNM included 175 of the 178 measured specimens, whereas the other included only three specimens from three different localities, which did not form a homogeneous cluster (Figure 3).The three specimens included in the second NNM can be considered outliers rather than representatives of a second species.The first three components of a PCA based on the seven shell parameters together explained 94% of the variance, whereas the following three components explained only 6% in addition.However, the varsel procedure recommended discarding the first three components because they are not very useful for group discrimination.A single NMM was favoured if we used the first three components of the PCA for fitting the NMMs.The results indicate that the variation in the dataset can best be described by a single NMM and that V. ambigua and V. piscinalis in the sense of Vinarski et al. [14] represent artificial subdivisions of the variability in a single species.

Molecular Phylogeny
The COI alignment was 649 base pairs (bp) long, and the 16S rDNA alignment was 453 bp, yielding a total of 1102 bp.The ModelFinder analysis suggested keeping the four initial partitions separate and using the following substitution models: TN+F+G4/GTR+F+G4 for the first codon positions of COI for the maximum likelihood analysis and Bayesian inference analysis, respectively; F81+F/F81+F for the second codon positions of COI; TIM2+F+G4/GTR+F+G4 for the third codon positions of COI; and TIM+F+G4/HKY+F+I+G4 for 16S rDNA.
In the maximum likelihood tree and the Bayesian inference tree based on the concatenated partial COI and 16S rDNA sequences (Figure 4), the newly sequenced specimens of the V. piscinalis complex from the Hamburg surroundings formed a clade with a sample identified as V. cf.piscinalis from Brandenburg in Germany.A Valvata cf.piscinalis from southern Russia represents the sister group to the specimens of the V. piscinalis complex from Germany in the Bayesian inference tree but not in the maximum likelihood tree.However, there is no statistical support for the relationships among the specimens of the V. piscinalis complex.A specimen identified as V. piscinalis by Dinapoli and Klussman-Kolb

Molecular Phylogeny
The COI alignment was 649 base pairs (bp) long, and the 16S rDNA alignment was 453 bp, yielding a total of 1102 bp.The ModelFinder analysis suggested keeping the four initial partitions separate and using the following substitution models: TN+F+G4/GTR+F+G4 for the first codon positions of COI for the maximum likelihood analysis and Bayesian inference analysis, respectively; F81+F/F81+F for the second codon positions of COI; TIM2+F+G4/GTR+F+G4 for the third codon positions of COI; and TIM+F+G4/HKY+F +I+G4 for 16S rDNA.
In the maximum likelihood tree and the Bayesian inference tree based on the concatenated partial COI and 16S rDNA sequences (Figure 4), the newly sequenced specimens of the V. piscinalis complex from the Hamburg surroundings formed a clade with a sample identified as V. cf.piscinalis from Brandenburg in Germany.A Valvata cf.piscinalis from southern Russia represents the sister group to the specimens of the V. piscinalis complex from Germany in the Bayesian inference tree but not in the maximum likelihood tree.However, there is no statistical support for the relationships among the specimens of the V. piscinalis complex.A specimen identified as V. piscinalis by Dinapoli and Klussman-Kolb [27] from Lake Prespa in North Macedonia (EED-Phy-K388) forms a clade with Valvata (Cincinna) stenotrema Poli ński, 1929 from Lake Ohrid and is likely closely related or conspecific with this species (Figure 4).There is neither statistical support for the relationships of the species within Valvata (Cincinna) Mörch, 1864 nor for the relationships between the (sub-)genera.
The genetic distances between the COI sequences of the V. piscinalis complex samples from Hamburg are low.The K2P distances between the COI sequences vary between 0.000 and 0.0252 with an average of 0.0106.For comparison, the average K2P distances between V. [27] from Lake Prespa in North Macedonia (EED-Phy-K388) forms a clade with Valvata (Cincinna) stenotrema Poliński, 1929 from Lake Ohrid and is likely closely related or conspecific with this species (Figure 4).There is neither statistical support for the relationships of the species within Valvata (Cincinna) Mörch, 1864 nor for the relationships between the (sub-)genera.Our focus is the relationships of specimens of the V. piscinalis complex and the question of whether these relationships are congruent with the supposed morphological differences between V. ambigua and V. piscinalis.We constructed a median-joining network based on the COI sequences of the V. piscinalis complex samples from Hamburg and plotted the states of two ratios, shell shape (D/H) and relative umbilicus width (U/D), which distinguish V. ambigua and V. piscinalis according to Vinarski et al. [14], on the network (Figure 5).The distribution of the character states was not correlated with the genetic relationships of the individuals.Our focus is the relationships of specimens of the V. piscinalis complex and the question of whether these relationships are congruent with the supposed morphological differences between V. ambigua and V. piscinalis.We constructed a median-joining network based on the COI sequences of the V. piscinalis complex samples from Hamburg and plotted the states of two ratios, shell shape (D/H) and relative umbilicus width (U/D), which distinguish V. ambigua and V. piscinalis according to Vinarski et al. [14], on the network (Figure 5).The distribution of the character states was not correlated with the genetic relationships of the individuals.Valvata (Cincinna) ambigua Westerlund, 1873, a taxon of the Valvata (Cincinna) piscinalis complex, has recently been recorded in several regions of Central Europe [13,14,37,38].It is difficult to identify Valvata specimens using the keys, which do not even provide the variability of shell measurements, and the simplified outline drawings of Anistratenko [8] and Starobogatov et al. [9].Therefore, the morphometric data and photographs of V. ambigua and V. piscinalis provided by Vinarski et al. [14] were important advances.Vinarski et al. [14] specified that the shells of V. ambigua differ from those of V. piscinalis by a lower    Valvata (Cincinna) ambigua Westerlund, 1873, a taxon of the Valvata (Cincinna) piscinalis complex, has recently been recorded in several regions of Central Europe [13,14,37,38].It is difficult to identify Valvata specimens using the keys, which do not even provide the variability of shell measurements, and the simplified outline drawings of Anistratenko [8] and Starobogatov et al. [9].Therefore, the morphometric data and photographs of V. ambigua and V. piscinalis provided by Vinarski et al. [14] were important advances.Vinarski et al. [14] specified that the shells of V. ambigua differ from those of V. piscinalis by a lower number of whorls, a larger shell width, a relatively lower spire, a concave tangential line of the spire (almost straight in V. piscinalis) and clearly wider umbilicus than in V. piscinalis.The morphometric analyses of Vinarski et al. [14] were based on specimens from the Hamburg region in Germany, where both taxa were previously reported to co-occur.
Our Valvata samples from the Hamburg region included forms representing V. ambigua and V. piscinalis according to the measurements provided by Vinarski et al. (Table 2) [14].Plots of the shell measurements showed continuous variation in the shell diameter (D) and shell height (H) (Figure 2a), as well as in the shell shape (D/H) and relative width of the umbilicus (U/D) (Figure 2b).The results were similar when considering only specimens from the Reitbrooker Sammelgraben, where both taxa were previously reported to co-occur without intermediate forms [13].The plots of the measurements provide no evidence that the samples represent two distinct taxa.This is consistent with the principal component analysis shown by Vinarski et al. (Figure 4) [14] based on their shell measurements, in which the two taxa they distinguished also did not form distinct clusters.It is not an argument for the distinctness of artificially delimited subgroups of a continuously varying taxon that there are significant differences between these subgroups (as shown by Vinarski et al. (Table 2) [14]; see also [18]).A clustering approach based on NMMs using our morphometric dataset indicated that the dataset could be split into two NNMs.However, the second NNM includes only 3 of the 178 measured specimens (Figure 3), which do not form a morphologically homogeneous cluster but are better considered outliers.Therefore, the clustering approach indicated that V. ambigua and V. piscinalis are artificial subsets of a single NMM.The three outliers might be teratological specimens, with anomalies, e.g., caused by parasites.
In fact, 16S rDNA sequences from tissue samples of some specimens of the V. piscinalis complex from Reitbrooker Sammelgraben in Hamburg-Bergedorf and from the Untenburger Schleusengraben in Hamburg-Harburg showed that these specimens were infested by the oligochaet Chaetogaster limnaei von Baer, 1827 (Naididae).Smythe et al. [39] and Mack et al. [40] showed that Chaetogaster limnaei is a complex that includes several cryptic species.The Chaetogaster limnaei complex has been reported from a wide range of freshwater snails and mussels [41,42] but was not known from valvatids until now.The sequences we retrieved from the V. piscinalis complex (GenBank accession no.PP992211) belong to Chaetogaster sp.23 in the sense of Mack et al. [40], corresponding to the European Chaetogaster limnaei in the strict sense.The C. limnaei clade has adapted to a mixed lifestyle.The specimens can live as ectosymbionts on molluscs and feed on a diet of invertebrates, ciliates and diatoms or as parasites, which exclusively subsist on host cells [39,40].Chaetogaster limnaei apparently lives as a parasite in Valvata, as we obtained the Chaetogaster sequences from tissue samples of the snail hosts.The feeding of Chaetogaster may affect the growth and, therefore, shell size and structure of Valvata.However, the specimens in which we detected Chaetogaster were morphologically inconspicuous and belonged to NMM 1, as shown in Figure 3.
We used DNA sequences of mitochondrial genes to further test the hypothesis that the morphologically variable samples of the V. piscinalis complex from Hamburg represent two distinct species and to investigate their relationships with other Valvata taxa.The low genetic distances between the COI sequences of the V. piscinalis complex samples from Hamburg (maximum K2P distance 0.0252) do not indicate that these samples represent different species.Plotting the character states of shell shape (D/H) and relative umbilicus width (U/D), which distinguish V. ambigua and V. piscinalis according to Vinarski et al. [14], on a median-joining network based on COI sequences (Figure 5) showed that these character states are not correlated with the genetic relationships of the studied individuals.
The analyses of the morphometric data of the studied specimens of the V. piscinalis complex revealed no distinct clusters, although they included forms corresponding to V. ambigua and V. piscinalis, as defined by Vinarski et al. [14].The morphometric characters used by Vinarski et al. [14] to distinguish V. ambigua and V. piscinalis showed continuous variation.The genetic analyses showed that these character states are not correlated with the genetic relationships of the studied individuals.Accordingly, we consider Valvata (Cincinna) ambigua Westerlund, 1873, as delimited by Vinarski et al. [14], to be synonymous with Valvata (Cincinna) piscinalis (Müller, 1774).Samples from other regions identified as V. ambigua or V. piscinalis should be analysed to confirm their conspecificity with V. piscinalis as delimited here.Samples from Gothenburg in Sweden, the type locality of V. ambigua, and from Frederiksdal in Denmark, the type locality of V. piscinalis, should be genetically analysed to ensure that the names have been correctly interpreted.

Phylogenetic Relationships
We investigated the relationships of the newly sequenced samples of V. piscinalis with previously sequenced Valvata taxa [27][28][29], for which both COI and 16S rDNA sequences were available.The specimens of the V. piscinalis complex from the Hamburg surroundings formed a clade with a specimen from Brandenburg in Germany, identified as V. cf.piscinalis by Clewing et al. [28] (Figure 4).A Valvata cf.piscinalis from southern Russia represents the sister group to the specimens of the V. piscinalis complex from Germany in the Bayesian inference tree but not in the maximum likelihood tree.A further specimen identified as V. piscinalis by Dinapoli and Klussman-Kolb [27] from Lake Prespa in North Macedonia forms a clade with Valvata (Cincinna) stenotrema Poli ński, 1929 from Lake Ohrid and is likely closely related or conspecific with this species (Figure 4).This illustrates the difficulties in the identification of the Valvata species resulting from few diagnostic shell characters.The lack of statistical support for the relationships between the Valvata (Cincinna) species as well as for the relationships between the subgenera indicates that more sequence data and nuclear sequence data are necessary for a robust reconstruction of the phylogeny of the Valvatidae.The polyphyly of the samples identified as V. confusa Westerlund, 1897, by Clewing et al. [28] and Saito et al. [29], the polyphyly of the samples identified as V. hokkaidoensis Miyadi, 1935 by Saito et al. [29] in the maximum likelihood tree (Figure 4) and the very low genetic distances between samples identified as V. confusa, V. brevicula Kozhov, 1936 (V.aliena Westerlund, 1876 according to the GenBank entries), Valvata sp.I and Valvata sp.II by Clewing et al. [28] show that integrative taxonomic studies are also necessary to clarify the delimitation of other Valvata species.

Supplementary Materials:
The following supporting information can be downloaded at https: //www.mdpi.com/article/10.3390/d16070419/s1,Table S1.Shell measurements and ratios of specimens of the Valvata piscinalis complex from northern Germany; Table S2: Data of specimens used for molecular genetics.

Figure 2 .
Figure 2. Plots of the shell parameters of specimens of the Valvata piscinalis complex from the Hamburg region and the Ra eburger See.(a) Shell height (H) versus shell diameter (D).(b) Relative umbilicus width (U/D) versus shell shape (D/H).Specimens from the Reitbrooker Sammelgraben in Hamburg-Bergedorf, where both species are said to co-occur syntopically [13], are highlighted in red.

Figure 2 .
Figure 2. Plots of the shell parameters of specimens of the Valvata piscinalis complex from the Hamburg region and the Ratzeburger See.(a) Shell height (H) versus shell diameter (D).(b) Relative umbilicus width (U/D) versus shell shape (D/H).Specimens from the Reitbrooker Sammelgraben in Hamburg-Bergedorf, where both species are said to co-occur syntopically [13], are highlighted in red.

Figure 2 .
Figure 2. Plots of the shell parameters of specimens of the Valvata piscinalis complex from the Hamburg region and the Ra eburger See.(a) Shell height (H) versus shell diameter (D).(b) Relative umbilicus width (U/D) versus shell shape (D/H).Specimens from the Reitbrooker Sammelgraben in Hamburg-Bergedorf, where both species are said to co-occur syntopically [13], are highlighted in red.

Figure 3 .
Figure 3.The two best supported NMMs in the space defined by the three principal components most useful for group discrimination (PC4−PC6).The ellipse shows the 95% high-density region for the normal distribution of morphological group 1.

Figure 3 .
Figure 3.The two best supported NMMs in the space defined by the three principal components most useful for group discrimination (PC4−PC6).The ellipse shows the 95% high-density region for the normal distribution of morphological group 1.

Figure 4 .
Figure 4. Maximum likelihood tree of Valvatidae based on the concatenated partial sequences of COI and 16S rDNA.Species names and extraction voucher numbers are given at the tips of the tree.Newly sequenced individuals are shown in red.Bootstrap support values from the maximum likelihood analysis ≥70 and posterior probabilities from the Bayesian inference analysis ≥0.95 are indicated at the branches.

Figure 4 .
Figure 4. Maximum likelihood tree of Valvatidae based on the concatenated partial sequences of COI and 16S rDNA.Species names and extraction voucher numbers are given at the tips of the tree.Newly sequenced individuals are shown in red.Bootstrap support values from the maximum likelihood analysis ≥70 and posterior probabilities from the Bayesian inference analysis ≥0.95 are indicated at the branches.

Figure 5 .
Figure 5. Median-joining network based on the partial COI sequences of specimens of the Valvata piscinalis complex from the Hamburg region (see Supplementary TableS2for a list of specimens and GenBank accession numbers).(a) Specimens were classified according to shell shape (D/H).(b) Specimens were classified according to relative umbilicus width (U/D).
Figure 5. Median-joining network based on the partial COI sequences of specimens of the Valvata piscinalis complex from the Hamburg region (see Supplementary TableS2for a list of specimens and GenBank accession numbers).(a) Specimens were classified according to shell shape (D/H).(b) Specimens were classified according to relative umbilicus width (U/D).
Figure 5. Median-joining network based on the partial COI sequences of specimens of the Valvata piscinalis complex from the Hamburg region (see Supplementary TableS2for a list of specimens and GenBank accession numbers).(a) Specimens were classified according to shell shape (D/H).(b) Specimens were classified according to relative umbilicus width (U/D).

Figure 5 .
Figure 5. Median-joining network based on the partial COI sequences of specimens of the Valvata piscinalis complex from the Hamburg region (see Supplementary TableS2for a list of specimens and GenBank accession numbers).(a) Specimens were classified according to shell shape (D/H).(b) Specimens were classified according to relative umbilicus width (U/D).
Figure 5. Median-joining network based on the partial COI sequences of specimens of the Valvata piscinalis complex from the Hamburg region (see Supplementary TableS2for a list of specimens and GenBank accession numbers).(a) Specimens were classified according to shell shape (D/H).(b) Specimens were classified according to relative umbilicus width (U/D).
Figure 5. Median-joining network based on the partial COI sequences of specimens of the Valvata piscinalis complex from the Hamburg region (see Supplementary TableS2for a list of specimens and GenBank accession numbers).(a) Specimens were classified according to shell shape (D/H).(b) Specimens were classified according to relative umbilicus width (U/D).